src/vsp/llm/llm_service.py

from abc import ABC, abstractmethod


class RateLimitError(Exception):
    """Custom error for handling rate limit errors from the OpenAI API."""

    pass


class LLMService(ABC):
    """
    Abstract base class for Language Model Services.

    This class defines the interface for interacting with various language model services.
    Concrete implementations should provide the specific logic for invoking the language model.

    Methods:
        invoke(user_prompt: str | None = None,
               system_prompt: str | None = None,
               partial_assistant_prompt: str | None = None,
               max_tokens: int = 1000,
               temperature: float = 0.0) -> str | None:
            Abstract method to invoke the language model with given prompts and parameters.
    """

    @abstractmethod
    async def invoke(
        self,
        user_prompt: str | None = None,
        system_prompt: str | None = None,
        partial_assistant_prompt: str | None = None,
        max_tokens: int = 1000,
        temperature: float = 0.0,
    ) -> str | None:
        """
        Invoke the language model with the given prompts and parameters.

        Args:
            user_prompt (str | None): The main prompt from the user.
            system_prompt (str | None): A system message to set the context.
            partial_assistant_prompt (str | None): A partial response from the assistant.
            max_tokens (int): Maximum number of tokens in the response.
            temperature (float): Sampling temperature for response generation.

        Returns:
            str | None: The generated response from the language model, or None if no response.
        """
        pass